Amazon.com Inc. is getting the year off to a rolling start with a flurry of improvements to its public cloud aimed at filling in more of the logistical details that need addressing to support enterprise workloads. At the tip of the spear is a new connectivity option that enables regular virtual machines to access resources located in dedicated network partitions more directly than before.
ClassicLink extends the controls available within the isolated chunks of Amazon’s infrastructure that organizations carve out for their most sensitive applications to the public section, where administrators can now spin up instances with comparable security to their logical safety zones. The main intent is to make it easier for customers with existing deployments on the shared network to migrate into dedicated partitions.
In addition to helping practitioners move between private and public infrastructure, Amazon is also taking some of the hassle out of shifting across accounts, a chore for organizations that separate their environments for security and billing reasons. The company has integrated its identity and access management service into the management console so that it’s possible to switch among profiles with a few clicks instead of having to log out and input different credentials every time.
To help admins in the public sector be as productive as their private sector peers, Amazon is making several of the services that have previously only been accessible to private organizations available on its government-specific environment, including CloudTrail. The free utility automatically records usage information and periodically delivers the records to the user, which is especially handy for tracking resource consumption and other activity across multiple accounts.
The update also extends that alerting capability to Spot Instances, a service that allows users to specify a price at which they’re willing to purchase spare computational capacity and only have their instances run when the rate is below the threshold. Amazon now issues a two-minute notice before a Spot Instance is terminated, which should be just enough for an application to save its state, upload final log files and shut off in an orderly fashion.
Rounding out the upgrade is a new native implementation of the Java Message Service (JVM) that can handle communications among cloud applications written in that language. It offers a more convenient alternative to setting up a JVM cluster from scratch and managing it as the deployment changes, which is the only option that customers had up until now.